Tài liệu Báo cáo khoa học: Mutational analysis of plasminogen activator inhibitor-1 Interactions of a-helix F and its neighbouring structural elements regulates the activity and the rate of latency transition pdf

Mutational analysis of plasminogen activator inhibitor-1
Interactions of a-helix F and its neighbouring structural elements regulates
the activity and the rate of latency transition
Troels Wind*, Jan K. Jensen, Daniel M. Dupont, Paulina Kulig and Peter A. Andreasen
Laboratory of Cellular Protein Science, Department of Molecular Biology, Aarhus University, Denmark
The serpin plasminogen activator inhibitor-1 (PAI-1) is a
fast and specific inhibitor of the plasminogen activating
serine proteases tissue-type and urokinase-type plasminogen
activator and, as such, an important regulator in turnover of
extracellular matrix and in fibrinolysis. PAI-1 spontaneously
loses its antiproteolytic activity by inserting its reactive centre
loop (RCL) as strand 4 in b-sheet A, thereby converting to
the so-called latent state. We have investigated the import-
ance of the amino acid sequence of a-helix F (hF) and the
connecting loop to s3A (hF/s3A-loop) for the rate of latency
transition. We grafted regions of the hF/s3A-loop from
antithrombin III and a
1
-protease inhibitor onto PAI-1,
creating eight variants, and found that one of these rever-
sions towards the serpin consensus decreased the rate of
latency transition. We prepared 28 PAI-1 variants with
individual residues in hF and b-sheet A replaced by an
alanine. We found that mutating serpin consensus residues
always had functional consequences whereas mutating
nonconserved residues only had so in one case. Two variants
had low but stable inhibitory activity and a pronounced
tendency towards substrate behaviour, suggesting that
insertion of the RCL is held back during latency transition as
well as during complex formation with target proteases. The
data presented identify new determinants of PAI-1 latency

transition and provide general insight into the characteristic
loop–sheet interactions in serpins.
Keywords: alignment; conformation; mutational analysis;
PAI-1; proteases; serpin.
Plasminogen activator inhibitor-1 (PAI-1) is the primary
inhibitor of both urokinase-type and tissue-type plasmino-
gen activator (uPA and tPA, respectively) and as such is an
important regulator of physiological events in which
plasmin-catalysed extracellular proteolysis is involved.
PAI-1 belongs to the serine protease inhibitor (serpin)
family whose antiproteolytic activity is governed by their
structural flexibility. In the active serpin conformation, the
reactive centre loop (RCL) with the P
1
–P
1
¢ bait peptide
bond is surface exposed. Formation of the covalent serpin–
protease complex involves a Michaelis docking complex,
cleavage of the P
1
–P
1
¢ peptide bond, linkage of the active
site Ser of the protease to the carboxyl group of P
1
by an
ester bond and insertion of the N-terminal part of the RCL
as strand 4 in b-sheet A (s4A) of the serpin. Consequently,
the protease is trapped in a covalent acyl-enzyme complex in

which its reactive site is distorted, as illustrated by the crystal
structure of the complex between a
1
-protease inhibitor
(a
1
PI,alsoreferredtoasa
1
-antitrypsin) and trypsin [1].
Under some conditions, however, RCL insertion is delayed,
resulting in hydrolysis of the ester bond, release of free
protease and insertion of the cleaved RCL as s4A. This
pathway is referred to as substrate behaviour of the serpin.
Complex formation between serpins and their cognate
proteases is fuelled by the thermodynamic properties of the
serpin. Accordingly, insertion of the RCL as s4A and
the ensuing structural rearrangements of the serpin stabilizes
the molecule in a so-called ÔrelaxedÕ conformation, as
opposed to the metastable ÔstressedÕ conformation with the
RCL exposed on the surface (reviewed in [2–4]).
PAI-1 spontaneously converts into a relaxed conforma-
tion at a significant rate without cleavage of the RCL (for a
review see [5]). During this structural transformation,
referred to as latency transition, the N-terminal part of
the intact RCL is inserted as s4A [6] (Fig. 1). Latent versions
of the serpins antithrombin III (ATIII) [7], a
1
-protease
inhibitor (a
1

PI) [8], and a
1
-antichymotrypsin (a
1
ACT) [9]
have also been isolated, but none of these undergo this
transition as readily as PAI-1. The physiological role of
PAI-1 latency transition, if any, remains elusive [5].
Some PAI-1 variants with single mutations and modest
decreases in the rate of latency transition have been
obtained through heuristic protein engineering [10,11] while
others have been identified by chance [12–15]. The variants
with the slowest latency transition carry multiple mutations
Correspondence to J. K. Jensen, Laboratory of Cellular Protein
Science, Department of Molecular Biology, Aarhus University,
Gustav Wieds Vej 10C, 8000 A
˚
rhus C, Denmark.
Fax: + 45 86123178, Tel.: + 45 89425074,
E-mail:
Abbreviations: PAI-1, plasminogen activator inhibitor-1; RCL,
reactive centre loop; a
1
PI, a
1
-protease inhibitor (a
1
-antitrypsin);
a
1

ACT, a
1
-antichymotrypsin; ATIII, antithrombin III;
hF, a-helix F; HMK, heart muscle kinase.
Enzyme: uPA, urokinase-type plasminogen activator (EC 3.4.21.73).
*Present address: Centre for Vascular Research, School of Medical
Sciences, The University of New South Wales,
Sydney NSW 2052, Australia.
(Received 4 December 2002, revised 7 February 2003,
accepted 13 February 2003)
Eur. J. Biochem. 270, 1680–1688 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03524.x
and have been obtained by random mutagenesis followed
by screening or selection procedures. Berkenpas et al.
isolated several PAI-1 variants with increased stability of
whichthemoststable,referredtointhefollowingasPAI-
1
stab
, carried four amino acid substitutions (N152H, K156T,
Q321L and M356I) [16]. The three-dimensional structure of
PAI-1
stab
has been determined and reveals that the stabil-
izing amino acid substitutions N152H and K156T induces a
3
10
-helix spanning residues 155–157 in the loop connecting
a-helix F to b-strand 3A (the hF/s3A-loop, Fig. 1) [17–19].
Likewise, Stoop et al. isolated a panel of stable PAI-1
variants of which the most stable carried 10 amino acid
substitutions [20]. The structure of this variant, however, has

not been determined. A prevalent theme among the
stabilizing amino acid substitutions is a reversion towards
the consensus sequence of inhibitory serpins, suggesting that
part of the molecular basis for latency transition can be
found in regions of PAI-1 that deviate from this consensus
[16,20].
Based on the following observations, we speculated that
the amino acid composition of the hF/s3A-loop and hF
plays a role in the stressed-to-relaxed transition of PAI-1:
The hF/s3A-loop must fold away from b-sheet A during the
structural transition to allow insertion of the RCL as s4A [6]
(Fig. 1). Stabilizing mutations have been identified in the
hF/s3A-loop [16,20] and the aforementioned 3
10
-helix in
PA1–1
stab
has been suggested to, at least in part, be
responsible for the increased stability of this variant [17–19].
Some monoclonal antibodies towards hF and the hF/s3A-
loop [21–24] as well as deletion of this region [25] induce
substrate behaviour of PAI-1. In addition, Gettins recently
hypothesized that a thermodynamically unfavourable dis-
location of hF, resulting from insertion of the RCL,
provides an energy-reservoir that subsequently fuels the
crushing of the protease concomitantly with the return of
hF to its normal position [26].
In the present study, we used structural alignments to
define residues in the hF/s3A-loop in PAI-1 that deviate
from the serpin consensus and replaced them with the

corresponding residues from the representative inhibitory
serpins a
1
PI and ATIII. As there are no discrepancies
between the PAI-1 sequence and the serpin consensus in
hF, we chose alanine-scanning mutagenesis as a means of
identifying hF-residues with importance for RCL insertion.
Likewise, residues from b-sheet A with putative contacts to
hF or the hF/s3A-loop were individually replaced by
alanine. In total, 38 PAI-1 variants were characterized in
terms of latency transition and functional behaviour upon
interaction with uPA and several side chains involved in
the retardation of latency transition, and hence of
importance for the functional stability of stressed PAI-1,
were identified. Finally, we describe PAI-1 variants that
can adapt remarkably stable conformations and predom-
inantly behave as substrates for uPA. As will be discussed,
the data presented can probably be extrapolated to other
serpins and thus provide further insight into the molecular
details of the stressed-to-relaxed transition of these
proteins [27].
Fig. 1. Ribbon diagrams of relaxed, latent PAI-1 (right) and stressed, active PAI-1
stab
(left) [18]. Insertion of the reactive centre loop (red) as strand 4
in b-sheet A (pink) requires mobility of hF and the hF/s3A-loop (orange).
Ó FEBS 2003 Mutational analysis of PAI-1 (Eur. J. Biochem. 270) 1681
Materials and methods
Cloning, mutagenensis and purification of PAI-1
The cDNA for human PAI-1 was modified to include a
N-terminal His

6
-tag plus a recognition motif for heart
muscle kinase and cloned into the Escherichia coli expres-
sion vector pT7-PL [28]. Mutations in PAI-1 were intro-
duced with the QuickChange Site-Directed Mutagenesis Kit
(Stratagene) as described by the manufacturer, except that
the final product was electroporated into E. coli DH5a cells,
and confirmed by sequencing using the Thermo Sequenase
II Dye Terminator Cycle Sequencing Kit (Amersham
Pharmacia Biotech) and a 373A ABI sequencer (Applied
Biosystems). Numbering of residues in PAI-1 was S
1
-A
2
-V
3
-
H
4
-H
5
… [29].
For gene expression, individual colonies of transformed
E. coli cells BL21(DE3)pLysS (Novagen) were inoculated
into 2 · TY broth (16 gÆL
)1
tryptone, 10 gÆL
)1
yeast
extract, 5 gÆL

)1
NaCl) supplemented with 100 lgÆmL
)1
ampicillin and 34 lgÆmL
)1
chloramphenicol and incubated
overnight at room temperature. The cultures were diluted
1 : 20 and incubated at 37 °CuntilaD
600
between 0.7 and
0.9. A final concentration of 0.5 m
M
isopropyl thio-b-
D
-
galactoside was added to induce gene expression and the
incubation was continued for 2 h. From this point, protein
purification was performed at 4 °C. The cells were harvested
by centrifugation (7000 g, 20 min), resuspended in 35 mL
phosphate-buffered saline (137 m
M
NaCl, 2.7 m
M
KCl,
1.4 m
M
KH
2
PO
4

,4.3m
M
Na
2
HPO
4
) and opened with
sonication. The bacterial lysates were cleared by centrifu-
gation (15 000 g, 30 min) and filtration (0.22 lm), supple-
mented with 2
M
NaCl, 10 m
M
imidazole and 5% glycerol,
and applied to a 5-mL Ni-nitrilotriacetic acid column
(Qiagen) equilibrated in the same buffer. After extensive
washing with the equilibration-buffer, PAI-1 was eluted by
increasing the concentration of imidazole to 200 m
M
.The
eluted protein was subjected to gel filtration on a Superdex
75 column (1.6 · 60 cm, Amersham Pharmacia Biotech)
that had been equilibrated in Hepes-buffered saline (10 m
M
Hepes, 0.14
M
NaCl, pH 7.4 at 37 °C) supplemented with
5% glycerol and a final concentration of 1
M
NaCl.

Fractions containing PAI-1 were pooled, the concentration
determined from A
280
using the calculated extinction
coefficient 0.77 mLÆmg
)1
Æcm
)1
[30], and stored at )80 °C
until used. This procedure routinely gave 2–15 mg PAI-1
per litre of culture and N-terminal sequencing revealed that
the recombinant protein had the expected N terminus, i.e.
(M)GSMGSHHHHHHGS
RRASV
3
…, where the initi-
ating M in parentheses is missing. The phosphorylation site
for heart muscle kinase (HMK, underlined) allows radio-
active labelling of the molecule, a feature not used in
the present study. The reactivity of His- and HMK-tagged
PAI-1 in terms of uPA inhibition and vitronectin binding
was not affected by this modification [31].
SDS/PAGE analysis of functional behaviour
Reactions between recombinant PAI-1 (100 lgÆmL
)1
)and
uPA (200 lgÆmL
)1
, Wakamoto Pharmaceutical Company,
Tokyo, Japan) were performed in HBS at 37 °Cfor30min

and quenched by boiling in SDS sample buffer. The reaction
products were subjected to nonreducing SDS/PAGE in
11% acrylamide gels followed by staining with Coomassie
brilliant blue. For time-course experiments, PAI-1
(200 lgÆmL
)1
) was incubated in HBS at 37 °C for up to
24 h before reaction with uPA, followed by SDS/PAGE.
Band intensities were determined by scanning densitometry.
Determination of functional half-lives
The general buffer for the assay described was HBS (pH 7.4
at 37 °C) supplemented with 0.25% gelatine and unless
stated otherwise, all incubations were at 37 °C. PAI-1 was
incubated at a concentration of 20 lgÆmL
)1
and at various
time-points, aliquots were taken for preparation of a
twofold dilution series in a 96-well plate with 100 lL
PAI-1 per well in concentrations ranging from 20 to
0.0098 lgÆmL
)1
. Immediately thereafter, 100 lL
0.5 lgÆmL
)1
uPA (0.25 or 0.125 lgÆmL
)1
uPA for PAI-1
variants with activity below 20%) was added to each well
followed by incubation for at least 5 min to allow complex
formation between uPA and PAI-1. The remaining uPA

activity in each well was determined as the absorbance
at 405 nm after addition of 25 lL0.3mgÆmL
)1
S-2444
(Chromogenix, Sweden) and further incubation for 40 min.
The specific inhibitory activity of PAI-1 at the various time-
points, i.e. the fraction of the total amount of PAI-1 forming
a stable complex with uPA, was calculated from the amount
of PAI-1 required to inhibit half the uPA. The half-life of
PAI-1 was finally calculated from an exponential decay plot
of the data obtained. Generally, only one preparation of
each PAI-1 variant was investigated, but the following were
investigated with two independent preparations, giving
indistinguishable results: wild-type, PAI-1(T96A), PAI-
1(F100A), PAI-1(V126A), PAI-1(F128A), PAI-1(I137A),
PAI-1(I138A), PAI-1(N139A), PAI-1(W141A), PAI-
1(T146A) and PAI-1(M149K).
Structural analysis
Structural analysis was based on the following depositions
in the Protein Data Bank ([32]; PDB-ID is given in
parenthesis): active (1DVM) and latent (1DVN) PAI-1
[18], active a
1
-PI (1QLP) [33], cleaved a
1
-PI (7API) [34],
active and latent antithrombin (1E05) [35].
SWISS PDB
-
VIEWER

v3.51 ( was used for visu-
alization and structural alignments.
Statistical analysis
Rates of latency transition were compared using an
unpaired t-test.
Results
Reversions to the serpin consensus in the hF/s3A-loop
of PAI-1
To determine if the hF/s3A-loop governs latency transition
of PAI-1 by more readily allowing insertion of the intact
RCL than the corresponding loop from other inhibitory
serpins, we prepared PAI-1 variants with hF/s3A-loops that
mimic those found in the inhibitory serpins a
1
PI and ATIII,
1682 T. Wind et al. (Eur. J. Biochem. 270) Ó FEBS 2003
respectively. Table 1 shows a structure-based sequence
alignment of hF/s3A-loops from the relaxed serpin
structures cleaved a
1
PI [34], latent PAI-1 [18], and latent
ATIII [35]. Also, the consensus serpin sequence of the
hF/s3A-loop is included in Table 1 (adapted from [2]).
Alignment of the three structures was performed accord-
ing to the C
a
atoms in the rigid serpin fragment 2c [36],
encompassing residues 129–155 in PAI-1. The hF/s3A-
loop of PAI-1 deviates from the consensus serpin
sequence at position 149 (M instead of a basic residue,

e.g. K168 in a
1
PI) and position 152 (Asn instead of an
acidic residue, e.g. D171 in a
1
PI). Also, the stretch
GKGA(155–158) appears more flexible in PAI-1 than the
corresponding stretch in most other serpins, either because
of its length (e.g. compared to a
1
PI) or the lack of Pro
residues (e.g. compared to ATIII) (Table 1 and [2]). We
prepared the PAI-1 variants PAI-1(M149K), PAI-
1(N152D), PAI-1(G155K, D(156–157), A158E) and PAI-
1(G155P, K156S, G157E) where the latter two have the
stretch between position 155 and 158 replaced by the
corresponding stretch from a
1
PI and ATIII, respectively
(Table 1). These variants were all found to be not
significantly different from the wild-type in terms of
specific inhibitory activity (Table 2). Thus, the introduced
mutations did not compromise the correct folding of PAI-1
in its active conformation.
Compared with the wild-type protein, PAI-1(M149K),
and to a lesser extent PAI-1(G155K, D(156–157), A158E),
had a decreased rate of latency transition; PAI-1(N152D)
had a similar rate; and a slightly increased rate for PAI-
1(G155P, K156S, G157E) was counteracted by introducing
the N152D mutation (Table 2).

The two mutations M149K and N152D were introduced
individually or together in the PAI-1(G155K, D(156–157),
A158E) background and the resulting variants were found
to behave as wild-type PAI-1 towards uPA in terms of
inhibitory activity (Table 2). Combining the M149K and
[G155K, D(156–157), A158E] mutations did not decrease
the rate of latency transition compared to M149K alone
(T
½
¼ 156 ± 13 min vs. 136 ± 24 min, P ¼ 0.16)
(Table 2).
Alanine scanning mutagenesis
The s5A residues K325 and K327 have been suggested
to coordinate a chloride ion between b-sheet A and the
hF/s3A-loop [18]. The s6A residues E283 and E285 are
potential partners for electrostatic interactions with K325
and K327, and E285 makes contact with the hF/s3A-loop in
PAI-1
stab
[18]. Finally, T96 (s2A) forms a hydrogen bond
to the hF-residues H145 in latent PAI-1 and W141 in
PAI-1
stab
[18] and F100 (s2A), V126 (s1A) and F128 (s1A/
hF-loop) form part of the hydrophobic interface between
hF and b-sheet A. These eight side chains (E283, E285,
K325, K327, T96, F100, V126 and F128), each of the
residues in hF (i.e. S129 to K147), and the hF/s3A-loop
residues M149 and N152 (see above) were substituted with
A and the resulting variants were characterized in terms of

Table 2. Reversions to the serpin-consensus in the hF/s3A-loop. For each PAI-1 variant, the specific inhibitory activity towards uPA was determined
in a peptidolytic assay and expressed as percentage of the theoretical maximum. The activity was monitored over time and the rate of latency
transition expressed as the functional half-life, t
½
. The averages and standard deviations for at least three independent experiments are given.
PAI-1 variant Activity (%) t
½
(min)
Wild-type 74 ± 13 63 ± 6
M149K 83 ± 11 136 ± 24*
N152D 99 ± 8 67 ± 3
G155P, K156S, G157E 75 ± 16 50 ± 2*
N152D, G155P, K156S, G157E 83 ± 12 67 ± 6
G155K, D(156–157), A158E 66 ± 12 73 ± 6*
M149K, G155K, D(156–157), A158E 58 ± 5 156 ± 13*
N152D, G155K, D(156–157), A158E 62 ± 8 76 ± 7*
M149K, N152D, G155K, D(156–157), A158E 93 ± 16 148 ± 17*
* Significantly different from the corresponding value for wild-type (P < 0.005).
Table 1. Structure-based sequence alignment of the hF/s3A-loops from the three inhibitory serpins PAI-1, a
1
PI and ATIII. Residue numbering is
according to PAI-1 [29]. The alignment is based on the three-dimensional structures of the relaxed conformations of the serpins (see text for details).
Also shown is the serpin consensus sequence for this region, adapted from [2].
Residue 147 148 149 150 151 152 153 154 155 156 157 158 159 160 170 171 172
Consensus – G K I – D
E
LL
V
––––V
L

I
D––T
PAI-1 K G M I S N L L G K G A V D Q L T
a
1
PI QGKI VDLVK– – ELDRDT
ATIII E G R I T D V I P S E A I N E L T
Ó FEBS 2003 Mutational analysis of PAI-1 (Eur. J. Biochem. 270) 1683
specific inhibitory activity, functional behaviour and the
rate of latency transition.
Among the variants tested, the following had a more than
threefold reduced specific inhibitory activity: F100A,
V126A, F128A, I137A, N139A, W141A, T146A, M149A,
N152A, and K327A (Table 3). Their functional behaviour
was analysed by treatment with uPA followed by SDS/
PAGE and scanning densitometry (Fig. 2). In that analysis,
inhibitory active PAI-1 will migrate as a complex with uPA
while PAI-1 exhibiting substrate behaviour will migrate
slightly faster than native PAI-1 due to cleavage of the
C-terminal 33 amino acids. Latent PAI-1 or PAI-1 other-
wise inert to uPA will comigrate with native PAI-1. This
analysis showed that the substitutions F100A, V126A,
F128A, I137A, N139A, W141A, T146A and N152A
increased the fraction of PAI-1 molecules behaving as a
substrate for uPA to between 40 and 60%, compared to
 15% for the wild-type protein (Fig. 2). They thus showed
a readily distinguishable substrate behaviour. To assay the
stability of this substrate behaviour, PAI-1 variants were
incubated for up to 24 h at 37 °C prior to reaction with uPA
and SDS/PAGE analysis. After 24 h, the fraction of

molecules behaving as a substrate for uPA decreased
approximately twofold for PAI-1(V126A), PAI-1(F100A),
PAI-1(F128A) and PAI-1(W141A) with a concomitant
increase in the fraction being inert to uPA. Substrate
behaviour remained almost constant for PAI-1(I137A),
PAI-1(N139A), PAI-1(T146A) and PAI-1(N152A) for 24 h
(Fig. 3 and not shown). The lower specific inhibitory
activity of M149A and K327A was associated solely with
an increased fraction in a form comigrating with native
PAI-1, and thus in an inert, probably latent conformation
(Fig. 2).
The rate of latency transition was determined for all
variants. Typical assays are depicted in Fig. 4, and the data
for all the variants are summarized in Table 3. Replacing
any of the residues E132, R135, D140, K147, M149, E283
and K327, respectively, with an A increased the rate of
latency transition more than twofold. Three variants, I137A,
V142A, and N152A, had a biphasic loss of activity, one
component with a significantly faster latency transition rate
and another component with a significantly slower latency
transition rate. The activity of the three variants PAI-1
(N139A), PAI-1(W141A) and PAI-1(T146A) remained
almost invariant for several hours at 37 °C(Table3).
The K325A substitution slightly delays latency transition
in PAI-1, which is in agreement with our previous obser-
vations [37,38]. Less pronounced, but still significantly
slower latency transition were observed with the substitu-
tions T96A and I138A. The E285A substitution also slightly
delays latency transition of wild-type PAI-1 whereas in the
PAI-1

stab
background, it accelerates latency transition.
Discussion
The only noteworthy stabilizing effect resulting from
reversions to the serpin consensus in the hF/s3A-loop was
seen for PAI-1(M149K) (Table 2). In the relaxed serpin
conformation, M149 in PAI-1 and the corresponding K168
Table 3. Alanine-scanning. For each PAI-1 variant, the specific
inhibitory activity towards uPA was determined in a peptidolytic assay
and expressed as percentage of the theoretical maximum. The activity
was monitored over time and the rate of latency transition expressed as
the functional half-life, t
½
. The averages and standard deviations for at
least three independent measurements are given for each variant.
PAI-1 variant Activity (%) t
½
(min)
Wild-type 74 ± 13 63 ± 6
T96A 91 ± 6 99 ± 23*
F100A 4 ± 1 54 ± 13
V126A 9 ± 3 42 ± 7*
F128A 2 ± 1 68 ± 2
S129A 80 ± 17 59 ± 8
E130A 59 ± 4 53 ± 3
V131A 74 ± 13 70 ± 4
E132A 51 ± 9 30 ± 1*
R133A 90 ± 10 57 ± 2
R135A 57 ± 6 26 ± 2*
F136A 88 ± 12 67 ± 5

I137A 18 ± 2 <10 and 170 ± 10*
a
I138A 52 ± 4 78 ± 4*
N139A 7 ± 1 335 ± 214*
D140A 63 ± 2 17 ± 0*
W141A 9 ± 2 431 ± 158*
V142A 32 ± 4 7 ± 2 and 120 ± 5*
a
K143A 88 ± 6 48 ± 3*
T144A 77 ± 13 47 ± 6*
H145A 75 ± 3 61 ± 7
T146A 7 ± 1 217 ± 33*
K147A 87 ± 12 26 ± 6*
M149A 10 ± 6 18 ± 2*
N152A 13 ± 2 17 ± 3 and 110 ± 21*
a
E283A 33 ± 3 14 ± 3*
E285A 97 ± 0 79 ± 3*
K325A 82 ± 19 106 ± 8*
K327A 6 ± 0 11 ± 1*
PAI-1
stab
69 ± 3 3340 ± 661*
PAI-1
stab
(E285A) 76 ± 3 680 ± 35*
,
**
a
A biphasic loss of activity was observed, suggesting a hetero-

geneity in the active fraction. Note that the wild-type residue at
position 134 is an A. * Significantly different from the corres-
ponding value for wild-type (P < 0.005); ** Significantly different
from the corresponding value for PAI-1
stab
(P < 0.005).
Fig. 2. Reaction products following reaction of PAI-1 variants with
uPA. PAI-1 variants (100 lgÆmL
)1
), indicated by their amino acid
substitution, were reacted with uPA (200 lgÆmL
)1
)at37°CinHBS,
pH 7.4 for 30 min and the products were separated by nonreducing
SDS/PAGE (11% acrylamide) followed by staining with Coomassie
brilliant blue. The migration of the uPA–PAI-1 complex, uPA, intact
(inert) PAI-1 and cleaved PAI-1 is indicated on the right.
1684 T. Wind et al. (Eur. J. Biochem. 270) Ó FEBS 2003
in a
1
PI are located at a critical position at the top of the hF/
s3A-loop right above the inserted s4A while in the stressed
conformation of PAI-1
stab
and a
1
PI, they stack against the
aromatic moiety of the s3A residues Y172 and F189,
respectively [18,33,34] (Fig. 5). Accordingly, we assume that
the aliphatic moiety of the introduced K in PAI-1(M149K)

allows the side chain to adapt the same orientation as the
original M and therefore the stabilizing effect of the M149K
substitution is likely to be governed by the introduced
positive charge. In stressed ATIII, the RCL is partially
inserted as the top of s4A and the equivalent of M149 from
PAI-1, i.e. R197, is located right above the bifurcation
between s3A and s5A [35]. In light of this, we suggest that a
positively charged side chain close to the point of initial
insertion of the RCL [36] represents an obstacle for the local
structural rearrangements required for the movements of
the RCL during latency transition. N152 is often replaced
by D in PAI-1 variants carrying several mutations that lead
to a decreased rate of latency transition [20,39]. However,
the N152D mutation does not per se delay latency transition
in PAI-1 (Table 2). Substitution of the stretch GKGA(155–
158) in PAI-1 with the corresponding stretch from a
1
PI or
ATIII had only modest effects on the rate of latency
transition (Table 2). Therefore, besides the M at position
149, deviations from the serpin consensus in the hF/s3A-
loop of PAI-1 does not contribute to the rate of latency
transition.
Alanine-scanning mutagenesis identified side chains that
contribute to the functional stability of PAI-1 as their
removal increased the rate of latency transition more than
twofold (Table 3). In principle, this observation can imply
two things: (a) the side chain in question contributes to the
thermodynamic stability of the stressed serpin conforma-
tion, which is why its removal makes the latency transition

energetically more favourable; (b) alternatively, the side
chain in question is instrumental in obstructing the
conformational changes occurring during the latency
transition. Accordingly, we propose that the hF side chains
of D140, K147 (forming a salt-bridge to s2A), M149
(packing against Y172 in s3A, Fig. 5), and the salt-bridge
between E132 and R135 [18] contribute to the thermo-
dynamic stability of the stressed conformation and/or
are important for the positioning of hF in a way
delaying the proper movements of the intact RCL during
latency transition. PAI-1(I137A), PAI-1(V142A) and
Fig. 3. Time-course experiment showing the substrate behaviour of selected variants. PAI-1 (200 lgÆmL
)1
)wasincubatedat37°C in HBS pH 7.4,
and at the indicated time points, aliquots were reacted with a twofold molar excess of uPA for 30 min. Reaction products were analysed by
nonreducing SDS/PAGE followed by staining with Coomassie brilliant blue. The migration of the uPA–PAI-1 complex, uPA, intact (inert) PAI-1
and cleaved PAI-1 is indicated on the right.
Fig. 4. Time-course experiment showing the inhibitory activity towards
uPA of representative PAI-1 variants as measured in a peptidolytic
assay. PAI-1 (20 lgÆmL
)1
) was incubated in HBS supplemented with
0.25% gelatine at 37 °C and at the indicated time-points, samples were
taken and the inhibitory activity determined. Activity was plotted
semilogarithmically against time. The experiment shown is a typical
one out of a total of at least three.
Ó FEBS 2003 Mutational analysis of PAI-1 (Eur. J. Biochem. 270) 1685
PAI-1 (N152A) all showed a biphasic loss of activity
suggesting a heterogeneity in the active fraction of these
variants. That substituting either of the juxtaposed residues

E283 (s6A) or K327 (s5A) with an A increases the rate of
latency transition may be related to the proposed role for
K327 in the coordination of a stabilizing chloride ion [18] or
suggest the existence of a stabilizing salt-bridge between the
two side chains (Fig. 5).
In contrast, substituting T96 in s2A, I138 in hF, K325 in
s5A or E285 in s6A with A increased the half-life of latency
transition by 24–68% (Table 3). The T96A substitution is a
reversion to the serpin consensus A/G (G115 in a
1
PI) [2],
suggesting that the absence of a side chain beyond the C
b
atom at this position increases the functional stability of the
stressed serpin. I138 is highly conserved among serpins (I157
in a
1
PI) [2], buried between hF and b-sheet A (Fig. 5) and
may be instrumental in promoting the translocation of hF
during RCL insertion. K325 is also conserved among
serpins (K335 in a
1
PI) [2] and its substitution for A in a
1
PI,
a
1
ACT and ATIII has been suggested to stabilize the
stressed conformation of these serpins by relieving the strain
of side chain overpacking between the K325 side chain and

residues in the hF/s3A-loop, i.e. the conserved I150 and
L153 in PAI-1 ([40,41], see Table 1). This is in good
agreement with our observation of a decreased rate of
latency transition for PAI-1(K325A) (Table 3 and [37,38]).
The side chains of K325 and E285 are juxtaposed, which is
why the E285A substitution may provide a spatial relief
mimicking the effect of the K325A substitution (Fig. 5). Of
note is that the side chain of E285 forms a hydrogen bond to
the backbone of the hF/s3A-loop in PAI-1
stab
[18] (Fig. 5)
and in contrast with the wild-type protein, the functional
stability of this variant is decreased by the E285A substi-
tution (Table 3). This advocates that the contact between
the E285 side chain and the hF/s3A-loop contributes to the
functional stability of PAI-1
stab
and that a similar contact is
not present in the stressed conformation of the wild-type
protein.
Inhibitory activity and substrate behaviour of PAI-
1(N139A) and PAI-1(T146A) were found to be invariant
for several hours (Fig. 3 and Table 3). Both events require
an exposed RCL, and it therefore seems that insertion of the
intact RCL during latency transition as well as insertion of
the cleaved RCL during complex formation with uPA is
retarded in these variants [42]. Both N139 and T146
are highly conserved among serpins (N158 and T165,
respectively, in a
1

PI) [2] and form hydrogen bonds to the
hF/s3A-loop [18] (Fig. 5). Considering the almost identical
phenotypes of PAI-1(N139A), PAI-1(T146A), and the close
spatial proximity and similar structural role of N139 and
T146, we find it likely that the structures of these variants,
with the RCL exposed, are similar and contains a distorted
hF that delays insertion of the RCL.
Substituting W141 with an A leads to substrate
behaviour and a low, stable inhibitory activity (Fig. 2
and Table 3). This W is located in the cleft between hF
and s2A (Fig. 5), and the presence of an aromatic side
chain at this position is common in serpins [2]. Mutation
of the equivalent Y160 in a
1
PI to A or W resulted in
decreased or increased thermodynamic stability, respect-
ively, and in line with our observation for PAI-1(W141A),
a
1
PI(Y160A) displayed a marked increase in substrate
behaviour [43].
Alanine substitution of the residues F100, V126, F128
and I137, respectively, led to increased substrate beha-
viour and a low unstable inhibitory activity, and for
I137A a biphasic loss of activity (Fig. 2 and Table 3). The
phenotype of these substitutions is therefore different from
that of N139A, T146A and W141A. The F residues at
positions 100 and 128 are highly conserved among serpins
(F119 and F147, respectively, in a
1

PI) [2] and buried in
the hydrophobic interface between hF and b-sheet A
(Fig. 5). The substrate behaviour and low activity, which
could not be increased by refolding in vitro (not shown),
of these variants suggest that these residues are pivotal for
Fig. 5. Selected residues important for the interactions between hF, the
hF/s3A-loop and b-sheet A, as seen in the structure of PAI-1
stab
[18] (for
an overview, see Fig. 1). H-bonds are indicated in green, a-helix F in
orange and parts of b-sheet A in pink. Backbone atoms of the s1A/hF-
loop (residues D127 to E130) and of the top of hF and the hF/s3A-
loop (residues T146 to L154) are shown in CPK colours. The following
side chains are shown in CPK colours and numbered (the secondary
structure element is given in parenthesis): 1, T96 (s2A); 2, F100 (s2A);
3, V126 (s1A); 4, F128 (s1A/hF-loop); 5, I137 (hF); 6, I138 (hF); 7,
N139 (hF); 8, W141 (hF); 9, V142 (hF); 10, T146 (hF); 11, M149 (hF/
s3A-loop); 12, Y172 (s3A); 13, E283 (s6A); 14, E285 (s6A); 15, K325
(s5A); 16, K327 (s5A). Note that in the structure of latent PAI-1, T96
forms a hydrogen bond with H145 from hF [18] (data not shown).
1686 T. Wind et al. (Eur. J. Biochem. 270) Ó FEBS 2003
the correct folding of the stressed serpin conformation.
V126 (conserved among serpins; V145 in a
1
PI [2]) and
I137 (not conserved) are partially exposed in the cleft
between hF and s1A (Fig. 5). The substrate behaviour of
the corresponding A-substituted variants suggests that
these residues are instrumental for the movements of hF
during complex formation.

As detailed above, several of the residues investigated in
this study are conserved among serpins and substituting any
of these with A changed the characteristics of PAI-1. In
addition, our observations for the M149A, M149K, N152A
and N152D variants of PAI-1 implicate the conserved side
chains K168 and D171 in a
1
PI (corresponding to M149 and
N152, respectively, in PAI-1) as contributors to the stability
of the stressed serpin conformation. The K335A substitu-
tion in a
1
PI, corresponding to the K325A substitution in
PAI-1, stabilizes a
1
PI by 6.5 kcalÆmol
)1
[40], suggesting that
the modest decrease in the rate of latency transition resulting
from the K325A substitution (Table 3) could reflect a
substantial stabilization of the stressed PAI-1 conformation.
We cannot, however, exclude the possibility that the
observed delay of latency transition resulting from amino
acid substitutions reflects features of PAI-1 not shared by
other serpins.
In contrast, none of the residues in hF that were
replaceable by A without functional consequences (i.e.
S129, E130, V131, R133, F136, K143, T144 and H145) are
conserved among serpins [2]. This supports the general
notion that conservation of a residue indicates its import-

ance for protein function. Furthermore, with the exception
of H145, these residues are pointing away from the hF/s3A-
loop and b-sheet A, suggesting that functionally important
residues should be sought in the interfaces between secon-
dary structural elements.
Conclusively, through mutagenesis we have now provi-
ded further evidence that the positioning of hF and its
movements relative to b-sheet A helps regulate the stressed-
to-relaxed transition of serpins in general and latency
transition of PAI-1 in particular. The data presented
provide novel insights into the determinants of serpin
stability located in and around hF and support the presence
of a novel serpin conformation that, due to rearrangements
in the top of hF, has a considerably delayed rate of RCL
insertion, both during latency transition and during com-
plex formation with uPA.
Acknowledgements
The excellent technical assistance of A. Christensen is gratefully
acknowledged. The work was supported by grants from the Danish
Cancer Society, The Danish Research Agency, and the Novo-Nordisk
Foundation.
References
1. Huntington, J.A., Read, R.J. & Carrell, R.W. (2000) Structure of a
serpin-protease complex shows inhibition by deformation. Nature
407, 923–926.
2. Irving, J.A., Pike, R.N., Lesk, A.M. & Whisstock, J.C. (2000)
Phylogeny of the serpin superfamily: implications of patterns of
amino acid conservation for structure and function. Genome Res.
10, 1845–1864.
3. Gils, A. & Declerck, P.J. (1998) Structure-function relationships in

serpins: current concepts and controversies. Thromb. Haemost. 80,
531–541.
4. Ye, S. & Goldsmith, E.J. (2001) Serpins and other covalent pro-
tease inhibitors. Curr. Opin. Struct. Biol. 11, 740–745.
5. Wind,T.,Hansen,M.,Jensen,J.K.&Andreasen,P.A.(2002)The
molecular basis for anti-proteolytic and non-proteolytic functions
of plasminogen activator inhibitor type-1. Roles of the reactive
centre loop, the shutter region, the flexible joint-region and the
small serpin fragment. Biol. Chem. 383, 21–36.
6. Mottonen, J., Strand, A., Symersky, J., Sweet, R.M., Danley,
D.E.,Geoghegan,K.F.,Gerard,R.D.&Goldsmith,E.J.(1992)
Structural basis of latency in plasminogen activator inhibitor-1.
Nature 355, 270–273.
7. Carrell, R.W., Stein, P.E., Fermi, G. & Wardell, M.R. (1994)
Biological implications of a 3 A
˚
structure of dimeric antithrombin.
Structure 2, 257–270.
8. Lomas, D.A., Elliott, P.R., Chang, W.S., Wardell, M.R. & Car-
rell, R.W. (1995) Preparation and characterization of latent
a
1
-antitrypsin. J. Biol. Chem. 270, 5282–5288.
9. Chang, W.S. & Lomas, D.A. (1998) Latent a
1
-antichymotrypsin.
A molecular explanation for the inactivation of a
1
-antichymo-
trypsin in chronic bronchitis and emphysema. J. Biol. Chem. 273,

3695–3701.
10. Lawrence, D.A., Olson, S.T., Palaniappan, S. & Ginsburg, D.
(1994) Engineering plasminogen activator inhibitor 1 mutants
with increased functional stability. Biochemistry 33, 3643–3648.
11.Tucker,H.M.,Mottonen,J.,Goldsmith,E.J.&Gerard,R.D.
(1995) Engineering of plasminogen activator inhibitor-1 to reduce
the rate of latency transition. Nat. Struct. Biol. 2, 442–445.
12. Sui, G.C. & Wiman, B. (1998) Stability of plasminogen activator
inhibitor-1: role of tyrosine221. FEBS Lett. 423, 319–323.
13. Sui, G.C. & Wiman, B. (2000) The B b-sheet in the PAI-1 molecule
plays an important role for its stability. Thromb. Haemost. 83,
896–901.
14. Ma
˚
ngs,H.,Sui,G.C.&Wiman,B.(2000)PAI-1stability:therole
of histidine residues. FEBS Lett. 475, 192–196.
15. Ngo, T.H., Hoylaerts, M.F., Knockaert, I., Brouwers, E. &
Declerck, P.J. (2001) Identification of a target site in plasminogen
activator inhibitor-1 that allows neutralization of its inhibitory
properties concomitant with an allosteric up-regulation of its
antiadhesive properties. J. Biol. Chem. 276, 26243–26248.
16. Berkenpas, M.B., Lawrence, D.A. & Ginsburg, D. (1995) Mole-
cular evolution of plasminogen activator inhibitor-1 functional
stability. EMBO J. 14, 2969–2977.
17. Sharp, A.M., Stein, P.E., Pannu, N.S., Carrell, R.W., Berkenpas,
M.B., Ginsburg, D., Lawrence, D.A. & Read, R.J. (1999) The
active conformation of plasminogen activator inhibitor 1, a target
for drugs to control fibrinolysis and cell adhesion. Struct. Fold.
Des. 7, 111–118.
18. Stout,T.J.,Graham,H.,Buckley,D.I.&Matthews,D.J.(2000)

Structures of active and latent PAI-1: a possible stabilizing role for
chloride ions. Biochemistry 39, 8460–8469.
19. Nar, H., Bauer, M., Stassen, J.M., Lang, D., Gils, A. & Declerck,
P.J. (2000) Plasminogen activator inhibitor 1. Structure of the
native serpin, comparison to its other conformers and implications
for serpin inactivation. J. Mol. Biol. 297, 683–695.
20. Stoop, A.A., Eldering, E., Dafforn, T.R., Read, R.J. & Panne-
koek, H. (2001) Different structural requirements for plasminogen
activator inhibitor 1 (PAI-1) during latency transition and pro-
teinase inhibition as evidenced by phage-displayed hypermutated
PAI-1 libraries. J. Mol. Biol. 305, 773–783.
21. Komissarov, A.A., Declerck, P.J. & Shore, J.D. (2002) Mechan-
isms of conversion of plasminogen activator inhibitor 1 from a
suicide inhibitor to a substrate by monoclonal antibodies. J. Biol.
Chem. 277, 43858–43865.
Ó FEBS 2003 Mutational analysis of PAI-1 (Eur. J. Biochem. 270) 1687
22. Bijnens, A.P., Gils, A., Knockaert, I., Stassen, J.M. &Declerck, P.J.
(2000) Importance of the hinge region between a-helix F and the
main part of serpins, based upon identification of the epitope of
plasminogen activator inhibitor type 1 neutralizing antibodies.
J. Biol. Chem. 275, 6375–6380.
23. Schousboe, S.L., Egelund, R., Kirkegaard, T., Preissner, K.T.,
Rodenburg, K.W. & Andreasen, P.A. (2000) Vitronectin and
substitution of a b-strand 5A lysine residue potentiate activity-
neutralization of PA inhibitor-1 by monoclonal antibodies against
a-helix F. Thromb. Haemost. 83, 742–751.
24. Wind, T., Jensen, M.A. & Andreasen, P.A. (2001) Epitope map-
ping for four monoclonal antibodies against human plasminogen
activator inhibitor type-1. Implications for antibody-mediated
PAI-1-neutralization and vitronectin-binding. Eur. J. Biochem.

268, 1095–1106.
25. Vleugels, N., Gils, A., Bijnens, A.P., Knockaert, I. & Declerck,
P.J. (2000) The importance of helix F in plasminogen activator
inhibitor-1. Biochim. Biophys. Acta. 1476, 20–26.
26. Gettins, P.G. (2002) The F-helix of serpins plays an essential,
active role in the proteinase inhibition mechanism. FEBS Lett.
523, 2–6.
27. Carrell, R.W. & Stein, P.E. (1996) The biostructural pathology of
the serpins: critical function of sheet opening mechanism. Biol.
Chem. Hoppe Seyler 377, 1–17.
28. Christensen, J.H., Hansen, P.K., Lillelund, O. & Thøgersen, H.C.
(1991) Sequence-specific binding of the N-terminal three-finger
fragment of Xenopus transcription factor IIIA to the internal
control region of a 5S RNA gene. FEBS Lett. 281, 181–184.
29. Andreasen,P.A.,Riccio,A.,Welinder,K.G.,Douglas,R.,Sar-
torio,R.,Nielsen,L.S.,Oppenheimer,C.,Blasi,F.&Danø,K.
(1986) Plasminogen activator inhibitor type-1: reactive center and
amino-terminal heterogeneity determined by protein and cDNA
sequencing. FEBS Lett. 209, 213–218.
30. Gill, S.C. & von Hippel, P.H. (1989) Calculation of protein
extinction coefficients from amino acid sequence data. Anal.
Biochem. 182, 319–326.
31. Jensen, J.K., Wind, T. & Andreasen, P.A. (2002) The vitronectin
binding area of plasminogen activator inhibitor-1, mapped by
mutagenesis and protection against an inactivating organochem-
ical ligand. FEBS Lett. 521, 91–94.
32. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N.,
Weissig,H.,Shindyalov,I.N.&Bourne,P.E.(2000)TheProtein
Data Bank. Nucleic Acids Res. 28, 235–242.
33. Elliott, P.R., Pei, X.Y., Dafforn, T.R. & Lomas, D.A. (2000)

Topography of a 2.0 A
˚
structure of a
1
-antitrypsin reveals targets
for rational drug design to prevent conformational disease. Protein
Sci. 9, 1274–1281.
34. Engh,R.,Lobermann,H.,Schneider,M.,Wiegand,G.,Huber,R.
& Laurell, C.B. (1989) The S variant of human a
1
-antitrypsin,
structure and implications for function and metabolism. Protein
Eng. 2, 407–415.
35. Skinner, R., Abrahams, J.P., Whisstock, J.C., Lesk, A.M., Carrell,
R.W. & Wardell, M.R. (1997) The 2.6 A
˚
structure of antithrombin
indicates a conformational change at the heparin binding site.
J. Mol. Biol. 266, 601–609.
36. Whisstock, J.C., Skinner, R., Carrell, R.W. & Lesk, A.M. (2000)
Conformational changes in serpins. I. The native and cleaved
conformations of a
1
-antitrypsin. J. Mol. Biol. 296, 685–699.
37. Hansen, M., Busse, M.N. & Andreasen, P.A. (2001) Importance
of the amino-acid composition of the shutter region of plasmi-
nogen activator inhibitor-1 for its transitions to latent and sub-
strate forms. Eur. J. Biochem. 268, 6274–6283.
38. Jensen, S., Kirkegaard, T., Pedersen, K.E., Busse, M., Preissner,
K.T., Rodenburg, K.W. & Andreasen, P.A. (2002) The role of

b-strand 5A of plasminogen activator inhibitor-1 in regulation of
its latency transition and inhibitory activity by vitronectin.
Biochim. Biophys. Acta. 1597, 301–310.
39. Stoop, A.A., Jespers, L., Lasters, I., Eldering, E. & Pannekoek, H.
(2000) High-density mutagenesis by combined DNA shuffling and
phage display to assign essential amino acid residues in protein–
protein interactions: application to study structure–function of
plasminogen activation inhibitor 1 (PAI-I). J. Mol. Biol. 301,
1135–1147.
40. Im, H., Seo, E.J. & Yu, M.H. (1999) Metastability in the in-
hibitory mechanism of human a
1
-antitrypsin. J. Biol. Chem. 274,
11072–11077.
41. Im, H. & Yu, M.H. (2000) Role of Lys335 in the metastability and
function of inhibitory serpins. Protein Sci. 9, 934–941.
42. Lawrence, D.A., Olson, S.T., Muhammad, S., Day, D.E.,
Kvassman, J.O., Ginsburg, D. & Shore, J.D. (2000) Partitioning
of serpin-proteinase reactions between stable inhibition and sub-
strate cleavage is regulated by the rate of serpin reactive center
loop insertion into b-sheet A. J. Biol. Chem. 275, 5839–5844.
43. Cabrita, L.D., Whisstock, J.C. & Bottomley, S.P. (2002) Probing
the role of the F-helix in serpin stability through a single trypto-
phan substitution. Biochemistry. 41, 4575–4581.
1688 T. Wind et al. (Eur. J. Biochem. 270) Ó FEBS 2003